Mathml

configdoc/usage.xml

@@ -238,6 +238,11 @@
    <line>279</line>
   </file>
  </directive>
+ <directive id="HTML.MathML">
+  <file name="HTMLPurifier/HTMLModuleManager.php">
+   <line>282</line>
+  </file>
+ </directive>
  <directive id="Attr.IDBlacklist">
   <file name="HTMLPurifier/IDAccumulator.php">
    <line>27</line>

library/HTMLPurifier.includes.php

@@ -162,6 +162,7 @@
 require 'HTMLPurifier/HTMLModule/Image.php';
 require 'HTMLPurifier/HTMLModule/Legacy.php';
 require 'HTMLPurifier/HTMLModule/List.php';
+require 'HTMLPurifier/HTMLModule/MathML.php';
 require 'HTMLPurifier/HTMLModule/Name.php';
 require 'HTMLPurifier/HTMLModule/Nofollow.php';
 require 'HTMLPurifier/HTMLModule/NonXMLCommonAttributes.php';
@@ -191,6 +192,7 @@
 require 'HTMLPurifier/Injector/AutoParagraph.php';
 require 'HTMLPurifier/Injector/DisplayLinkURI.php';
 require 'HTMLPurifier/Injector/Linkify.php';
+require 'HTMLPurifier/Injector/MathMLSpaceNormalize.php';
 require 'HTMLPurifier/Injector/PurifierLinkify.php';
 require 'HTMLPurifier/Injector/RemoveEmpty.php';
 require 'HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php';

library/HTMLPurifier.safe-includes.php

@@ -156,6 +156,7 @@
 require_once $__dir . '/HTMLPurifier/HTMLModule/Image.php';
 require_once $__dir . '/HTMLPurifier/HTMLModule/Legacy.php';
 require_once $__dir . '/HTMLPurifier/HTMLModule/List.php';
+require_once $__dir . '/HTMLPurifier/HTMLModule/MathML.php';
 require_once $__dir . '/HTMLPurifier/HTMLModule/Name.php';
 require_once $__dir . '/HTMLPurifier/HTMLModule/Nofollow.php';
 require_once $__dir . '/HTMLPurifier/HTMLModule/NonXMLCommonAttributes.php';
@@ -185,6 +186,7 @@
 require_once $__dir . '/HTMLPurifier/Injector/AutoParagraph.php';
 require_once $__dir . '/HTMLPurifier/Injector/DisplayLinkURI.php';
 require_once $__dir . '/HTMLPurifier/Injector/Linkify.php';
+require_once $__dir . '/HTMLPurifier/Injector/MathMLSpaceNormalize.php';
 require_once $__dir . '/HTMLPurifier/Injector/PurifierLinkify.php';
 require_once $__dir . '/HTMLPurifier/Injector/RemoveEmpty.php';
 require_once $__dir . '/HTMLPurifier/Injector/RemoveSpansWithoutAttributes.php';

library/HTMLPurifier/AttrDef/MathML/Character.php

@@ -0,0 +1,25 @@
+<?php
+
+/**
+ * Validates the MathML attribute character.
+ */
+
+class HTMLPurifier_AttrDef_MathML_Character extends HTMLPurifier_AttrDef
+{
+
+    /**
+     * @param string $char
+     * @param HTMLPurifier_Config $config
+     * @param HTMLPurifier_Context $context
+     * @return bool|string
+     */
+    public function validate($char, $config, $context)
+    {
+        if (mb_strlen($char) == 1 && strpos(' \t\n\r', $char) === false) {
+            return $char;
+        } elseif (preg_match('/&((#[0-9]+)|(#x[0-9A-Fa-f]+)|([0-9A-Za-z]+));/', $char)) {
+            return $char;
+        }
+        return false;
+    }
+}

library/HTMLPurifier/AttrDef/MathML/Color.php

@@ -0,0 +1,44 @@
+<?php
+
+/**
+ * Validates Color as defined by MathML.
+ */
+class HTMLPurifier_AttrDef_MathML_Color extends HTMLPurifier_AttrDef
+{
+
+    // MathML 3 only accepts HTML4 color names + transparent
+    static protected $colornames = array(
+        'aqua',
+        'black',
+        'blue',
+        'fuchsia',
+        'gray',
+        'green',
+        'lime',
+        'maroon',
+        'navy',
+        'olive',
+        'purple',
+        'red',
+        'silver',
+        'teal',
+        'white',
+        'yellow',
+        'transparent'
+    );
+
+    public function validate($color, $config, $context)
+    {
+
+        $color = trim($color);
+
+        if (preg_match('/(#[0-9A-Fa-f]{3})|(#[0-9A-Fa-f]{6})/', $color) || in_array(strtolower($color), static::$colornames)) {
+            return $color;
+        }
+
+        return false;
+    }
+
+}
+
+// vim: et sw=4 sts=4

library/HTMLPurifier/AttrDef/MathML/ID.php

@@ -0,0 +1,77 @@
+<?php
+
+/**
+ * Validates the MathML attribute ID.
+ * @note This just checks that the ID is valid. It explicitly avoids checking
+ *       or adding to the ID Accumulator because the MathML 3 DTD makes it a
+ *       point to allow repeated IDs.
+ */
+
+class HTMLPurifier_AttrDef_MathML_ID extends HTMLPurifier_AttrDef
+{
+
+    /**
+     * @param string $id
+     * @param HTMLPurifier_Config $config
+     * @param HTMLPurifier_Context $context
+     * @return bool|string
+     */
+    public function validate($id, $config, $context)
+    {
+
+        $id = trim($id); // trim it first
+
+        if ($id === '') {
+            return false;
+        }
+
+        $prefix = $config->get('Attr.IDPrefix');
+        if ($prefix !== '') {
+            $prefix .= $config->get('Attr.IDPrefixLocal');
+            // prevent re-appending the prefix
+            if (strpos($id, $prefix) !== 0) {
+                $id = $prefix . $id;
+            }
+        } elseif ($config->get('Attr.IDPrefixLocal') !== '') {
+            trigger_error(
+                '%Attr.IDPrefixLocal cannot be used unless ' .
+                '%Attr.IDPrefix is set',
+                E_USER_WARNING
+            );
+        }
+
+        // we purposely avoid using regex, hopefully this is faster
+
+        if ($config->get('Attr.ID.HTML5') === true) {
+            if (preg_match('/[\t\n\x0b\x0c ]/', $id)) {
+                return false;
+            }
+        } else {
+            if (ctype_alpha($id)) {
+                // OK
+            } else {
+                if (!ctype_alpha(@$id[0])) {
+                    return false;
+                }
+                // primitive style of regexps, I suppose
+                $trim = trim(
+                    $id,
+                    'A..Za..z0..9:-._'
+                );
+                if ($trim !== '') {
+                    return false;
+                }
+            }
+        }
+
+        $regexp = $config->get('Attr.IDBlacklistRegexp');
+        if ($regexp && preg_match($regexp, $id)) {
+            return false;
+        }
+
+        // if no change was made to the ID, return the result
+        // else, return the new id if stripping whitespace made it
+        //     valid, or return false.
+        return $id;
+    }
+}

library/HTMLPurifier/AttrDef/MathML/Length.php

@@ -0,0 +1,36 @@
+<?php
+
+/**
+ * Validates the MathML type length (not to be confused with either HTML nor
+ * CSS's length).
+ */
+
+class HTMLPurifier_AttrDef_MathML_Length extends HTMLPurifier_AttrDef
+{
+
+    /**
+     * @param string $string
+     * @param HTMLPurifier_Config $config
+     * @param HTMLPurifier_Context $context
+     * @return bool|string
+     */
+    public function validate($string, $config, $context)
+    {
+        $string = $this->parseCDATA($string);
+
+        // Optimizations
+        if ($string === '') {
+            return false;
+        }
+        if ($string === '0') {
+            return '0';
+        }
+
+        $length = HTMLPurifier_MathMLLength::make($string);
+        if (!$length->isValid()) {
+            return false;
+        }
+
+        return $length->toString();
+    }
+}

library/HTMLPurifier/AttrTypes.php

@@ -50,10 +50,23 @@ public function __construct()
 
         // "proprietary" types
         $this->info['Class'] = new HTMLPurifier_AttrDef_HTML_Class();
+        $this->info['CSS'] = new HTMLPurifier_AttrDef_CSS();
 
         // number is really a positive integer (one or more digits)
         // FIXME: ^^ not always, see start and value of list items
         $this->info['Number']   = new HTMLPurifier_AttrDef_Integer(false, false, true);
+
+        // MathML types
+        $this->info['MathML_ID'] = new HTMLPurifier_AttrDef_MathML_ID();
+        $this->info['MathML_Length'] = new HTMLPurifier_AttrDef_MathML_Length();
+        $this->info['MathML_UnsignedInteger'] = new HTMLPurifier_AttrDef_Integer(false, true, true);
+        $this->info['MathML_PositiveInteger'] = new HTMLPurifier_AttrDef_Integer(false, false, true);
+        $this->info['MathML_Integer'] = new HTMLPurifier_AttrDef_Integer(true, true, true);
+        $this->info['MathML_UnsignedNumber'] = new HTMLPurifier_AttrDef_CSS_Number(true);
+        $this->info['MathML_Number'] = new HTMLPurifier_AttrDef_CSS_Number(false);
+        $this->info['MathML_Character'] = new HTMLPurifier_AttrDef_MathML_Character();
+        $this->info['MathML_Color'] = new HTMLPurifier_AttrDef_MathML_Color();
+
     }
 
     private static function makeEnum($in)

library/HTMLPurifier/ConfigSchema/schema/HTML.MathML.txt

@@ -0,0 +1,7 @@
+HTML.MathML
+TYPE: bool
+DEFAULT: false
+--DESCRIPTION--
+<p>
+    Enable MathML.</p>
+--# vim: et sw=4 sts=4